Search

Purpose: The aim is to assess GPT-4V's (OpenAI) diagnostic accuracy and its capability to identify glaucoma-related features compared to expert evaluations. Design: Evaluation of multimodal large language models for reviewing fundus images in glaucoma. Subjects: A total of 300 fundus images from 3 public datasets (ACRIMA, ORIGA, and RIM-One v3) that included 139 glaucomatous and 161 nonglaucomatous cases were analyzed. Methods: Preprocessing ensured each image was centered on the optic disc. GPT-4's vision-preview model (GPT-4V) assessed each image for various glaucoma-related criteria: image quality, image gradability, cup-to-disc ratio, peripapillary atrophy, disc hemorrhages, rim thinning (by quadrant and clock hour), glaucoma status, and estimated probability of glaucoma. Each image was analyzed twice by GPT-4V to evaluate consistency in its predictions. Two expert graders independently evaluated the same images using identical criteria. Comparisons between GPT-4V's assessments, expert evaluations, and dataset labels were made to determine accuracy, sensitivity, specificity, and Cohen kappa. Main Outcome Measures: The main parameters measured were the accuracy, sensitivity, specificity, and Cohen kappa of GPT-4V in detecting glaucoma compared with expert evaluations. Results: GPT-4V successfully provided glaucoma assessments for all 300 fundus images across the datasets, although approximately 35% required multiple prompt submissions. GPT-4V's overall accuracy in glaucoma detection was slightly lower (0.68, 0.70, and 0.81, respectively) than that of expert graders (0.78, 0.80, and 0.88, for expert grader 1 and 0.72, 0.78, and 0.87, for expert grader 2, respectively), across the ACRIMA, ORIGA, and RIM-ONE datasets. In Glaucoma detection, GPT-4V showed variable agreement by dataset and expert graders, with Cohen kappa values ranging from 0.08 to 0.72. In terms of feature detection, GPT-4V demonstrated high consistency (repeatability) in image gradability, with an agreement accuracy of ≥89% and substantial agreement in rim thinning and cup-to-disc ratio assessments, although kappas were generally lower than expert-to-expert agreement. Conclusions: GPT-4V shows promise as a tool in glaucoma screening and detection through fundus image analysis, demonstrating generally high agreement with expert evaluations of key diagnostic features, although agreement did vary substantially across datasets. Financial Disclosure(s): Proprietary or commercial disclosure may be found in the Footnotes and Disclosures at the end of this article.

This study aims to develop deep learning (DL) models to predict the retinal nerve fiber layer (RNFL) thickness changes in glaucoma, facilitating the early diagnosis and monitoring of disease progression. Using the longitudinal data from two glaucoma studies (Diagnostic Innovations in Glaucoma Study (DIGS) and African Descent and Glaucoma Evaluation Study (ADAGES)), we constructed models using optical coherence tomography (OCT) scans from 251 participants (437 eyes). The models were trained to predict the RNFL thickness at a future visit based on previous scans. We evaluated four models: linear regression (LR), support vector regression (SVR), gradient boosting regression (GBR), and a custom 1D convolutional neural network (CNN). The GBR model achieved the best performance in predicting pointwise RNFL thickness changes (MAE = 5.2 μm, R2 = 0.91), while the custom 1D CNN excelled in predicting changes to average global and sectoral RNFL thickness, providing greater resolution and outperforming the traditional models (MAEs from 2.0–4.2 μm, R2 from 0.94–0.98). Our custom models used a novel approach that incorporated longitudinal OCT imaging to achieve consistent performance across different demographics and disease severities, offering potential clinical decision support for glaucoma diagnosis. Patient-level data splitting enhances the evaluation robustness, while predicting detailed RNFL thickness provides a comprehensive understanding of the structural changes over time.

Background We previously demonstrated that retinal ischemic perivascular lesions (RIPLs), which are indicative of ischemia in the middle retina, may be a biomarker of ischemic cardiovascular disease. In this study, we sought to determine the relationship between RIPLs and atrial fibrillation, a common source of cardiac emboli. Methods and Results In this case‐control study, we identified individuals between the ages of 50 and 90 years who had undergone macular spectral domain optical coherence tomography imaging. Individuals with atrial fibrillation were identified, and age‐ and sex‐matched individuals from the same pool, but without a diagnosis of atrial fibrillation, were selected as controls. Spectral domain optical coherence tomography scans were reviewed by 3 independent and masked observers for presence of RIPLs. The relationship between RIPLs and atrial fibrillation was analyzed using multivariable logistic regression models. There were 106 and 91 subjects with and without atrial fibrillation, respectively. The percentage of subjects with RIPLs was higher in the atrial fibrillation group compared with the control group (57.5% versus 37.4%; P=0.005). After adjusting for age, sex, smoking history, hypertension, diabetes, coronary artery disease, carotid stenosis, stroke, and myocardial infarction, the presence of RIPLs was significantly associated with atrial fibrillation, with an odds ratio of 1.91 (95% CI, 1.01–3.59). Conclusions RIPLs are significantly associated with atrial fibrillation, independent of underlying ischemic heart disease or cardiovascular risk factors. This association may inform the diagnostic cardiovascular workup for individuals with RIPLs incidentally detected on optical coherence tomography scan of the macula.

A longitudinal ophthalmic dataset was used to investigate multi-modal machine learning (ML) models incorporating patient demographics and history, clinical measurements, optical coherence tomography (OCT), and visual field (VF) testing in predicting glaucoma surgical interventions. The cohort included 369 patients who underwent glaucoma surgery and 592 patients who did not undergo surgery. The data types used for prediction included patient demographics, history of systemic conditions, medication history, ophthalmic measurements, 24-2 VF results, and thickness measurements from OCT imaging. The ML models were trained to predict surgical interventions and evaluated on independent data collected at a separate study site. The models were evaluated based on their ability to predict surgeries at varying lengths of time prior to surgical intervention. The highest performing predictions achieved an AUC of 0.93, 0.92, and 0.93 in predicting surgical intervention at 1 year, 2 years, and 3 years, respectively. The models were also able to achieve high sensitivity (0.89, 0.77, 0.86 at 1, 2, and 3 years, respectively) and specificity (0.85, 0.90, and 0.91 at 1, 2, and 3 years, respectively) at an 0.80 level of precision. The multi-modal models trained on a combination of data types predicted surgical interventions with high accuracy up to three years prior to surgery and could provide an important tool to predict the need for glaucoma intervention.

Purpose: To compare the diagnostic accuracy and explainability of a Vision Transformer deep learning technique, Data-efficient image Transformer (DeiT), and ResNet-50, trained on fundus photographs from the Ocular Hypertension Treatment Study (OHTS) to detect primary open-angle glaucoma (POAG) and identify the salient areas of the photographs most important for each model’s decision-making process. Design: Evaluation of a diagnostic technology. Subjects, Participants, and Controls: Overall 66 715 photographs from 1636 OHTS participants and an additional 5 external datasets of 16 137 photographs of healthy and glaucoma eyes. Methods: Data-efficient image Transformer models were trained to detect 5 ground-truth OHTS POAG classifications: OHTS end point committee POAG determinations because of disc changes (model 1), visual field (VF) changes (model 2), or either disc or VF changes (model 3) and Reading Center determinations based on disc (model 4) and VFs (model 5). The best-performing DeiT models were compared with ResNet-50 models on OHTS and 5 external datasets. Main Outcome Measures: Diagnostic performance was compared using areas under the receiver operating characteristic curve (AUROC) and sensitivities at fixed specificities. The explainability of the DeiT and ResNet-50 models was compared by evaluating the attention maps derived directly from DeiT to 3 gradient-weighted class activation map strategies. Results: Compared with our best-performing ResNet-50 models, the DeiT models demonstrated similar performance on the OHTS test sets for all 5 ground-truth POAG labels; AUROC ranged from 0.82 (model 5) to 0.91 (model 1). Data-efficient image Transformer AUROC was consistently higher than ResNet-50 on the 5 external datasets. For example, AUROC for the main OHTS end point (model 3) was between 0.08 and 0.20 higher in the DeiT than ResNet-50 models. The saliency maps from the DeiT highlight localized areas of the neuroretinal rim, suggesting important rim features for classification. The same maps in the ResNet-50 models show a more diffuse, generalized distribution around the optic disc. Conclusions: Vision Transformers have the potential to improve generalizability and explainability in deep learning models, detecting eye disease and possibly other medical conditions that rely on imaging for clinical diagnosis and management.

The molecular underpinnings driving uveal melanoma (UM) progression are unknown. Here the authors show that loss of Polycomb Repressive Complex 1 triggers chromosomal instability, which promotes inflammatory signaling and migration in UM.

Abstract Uveal melanoma, the most common intraocular primary cancer in adults, is characterized by striking variability in metastatic tendencies. BAP1 deletion in the primary tumor is associated with uveal melanoma metastasis, but it cannot always be resolved by bulk DNA sequencing of heterogeneous tumors. Here, we show that assessment of BAP1 methylation is an accurate and readily clinically actionable assay to accurately identify high-risk uveal melanoma patients.

Purpose. To evaluate clinical outcome during 24 months follow-up between small incision lenticule extraction combined with cross-linking (SMILE Xtra) and small incision lenticule extraction (SMILE) only. Setting. Ophthalmology Division of San Rossore Medical Center, Pisa, Italy. Design. Retrospective comparative case series. Methods. The study comprised 70 eyes (35 patients); 40 eyes were corrected using SMILE and 30 eyes were corrected using SMILE Xtra using a low energy protocol. The outcomes were compared at 1, 6, 12, and 24 months postoperatively. Results. The mean spherical equivalent (SEQ) reduced from −7.18 ± 1.21 D to −0.01 ± 0.09 D in the SMILE group and from −6.20 ± 2.99 D to −0.04 ± 0.1 D postoperatively in SMILE Xtra (p0.05). At 1, 6, 12, and 24 months, there were no statistically significant differences between the SMILE and SMILE Xtra groups in logarithm of the minimum angle of resolution (logMAR) uncorrected distance visual acuity (UDVA), safety, and efficacy index (p>0.05). The mean average keratometry (K-avg) at 1, 6, 12, and 24 months after surgery did not shown any statistically significant difference between SMILE and SMILE Xtra group (p>0.05). The mean maximum keratometry (K-max) readings at 1, 6, 12, and 24 months were not statistically significant between SMILE and SMILE Xtra group (p>0.05). The preoperative mean thinnest point pachymetry (TTP) was 543.90 ± 22.85 μm in the SMILE group and 523.40 ± 37.01 μm in the SMILE Xtra group (p0.05). At 24 months, the TTP was 408.29 ± 38.75 μm for the SMILE group and 402.22 ± 37 μm for the SMILE Xtra group (p>0.05). In the preoperative period, the mean maximum posterior elevation (MPE) was 8.63 ± 4.35 μm for SMILE and 8.13 ± 2.54 μm for SMILE Xtra (p>0.05). After the surgical procedure, both groups showed a statistically significant increase of the MPE (p0.05). In the preoperative period, the means of the root mean square (RMS) of high-order aberration (HOA) were 0.08 ± 0.03 μm for the SMILE group and 0.08 ± 0.03 μm for the SMILE Xtra group (p>0.05). At 24 months, the RMS of HOA was 0.13 ± 0.07 μm for the SMILE group and 0.14 ± 0.07 μm for the SMILE Xtra group (p>0.05). In the preoperative period, the root mean square of coma aberration (RMS-Coma) aberration was 0.06 ± 0.09 μm for the SMILE group and 0.04 ± 0.03 μm for the SMILE Xtra group (p>0.05). At 24 months, the coma aberration of SMILE group was 0.12 ± 0.21 μm and 0.16 ± 0.25 μm for SMILE Xtra group (p>0.05). Conclusions. SMILE Xtra procedure is a safe and simple procedure that can be offered to patients with high corneal ectasia risk because there were no differences in the indices of ectasia compared to the group treated only with SMILE which has a low corneal ectatic risk.

Background: Cardiovascular disease is the leading cause of mortality and disability worldwide. A noninvasive test that can detect underlying cardiovascular disease has the potential to identify patients at risk prior to the occurrence of adverse cardiovascular events. We sought to determine whether an easily observed imaging finding indicative of retinal ischemia, which we term ‘retinal ischemic perivascular lesions’ (RIPLs), could serve as a biomarker for cardiovascular disease. Methods: We reviewed optical coherence tomography (OCT) scans of individuals, with no underlying retinal pathology, obtained at UC San Diego Health from July 2014 to July 2019. We identified 84 patients with documented cardiovascular disease and 76 healthy controls. OCT scans were assessed for evidence of RIPLs. In addition, the 10-year atherosclerotic cardiovascular disease (ASCVD) risk calculator was used to risk-stratify the subjects into four different categories. Findings: Patients with documented cardiovascular disease had higher number of RIPLs compared to healthy controls (2.8 vs 0.8, p 37). The number of RIPLs in individuals with intermediate and high 10-year ASCVD risk scores was higher than in those with low ASCVD risk scores (1.7 vs 0.64, p = 0.02 and 2.9 vs 0.64, p 0.002, respectively). Interpretation: The presence of RIPLs, which are anatomical markers of prior retinal ischemic infarcts, is suggestive of coexisting cardiovascular disease. RIPLs detection, obtained from routine retinal scans, may thus provide an additional biomarker to identify patients at risk of developing adverse cardiovascular events. Funding: None.

The variational Bayesian independent component analysis-mixture model (VIM), an unsupervised machine-learning classifier, was used to automatically separate Matrix Frequency Doubling Technology (FDT) perimetry data into clusters of healthy and glaucomatous eyes, and to identify axes representing statistically independent patterns of defect in the glaucoma clusters.FDT measurements were obtained from 1,190 eyes with normal FDT results and 786 eyes with abnormal FDT results from the UCSD-based Diagnostic Innovations in Glaucoma Study (DIGS) and African Descent and Glaucoma Evaluation Study (ADAGES). For all eyes, VIM input was 52 threshold test points from the 24-2 test pattern, plus age.FDT mean deviation was -1.00 dB (S.D. = 2.80 dB) and -5.57 dB (S.D. = 5.09 dB) in FDT-normal eyes and FDT-abnormal eyes, respectively (p